| If we step back from the anecdotal arguments in this thread and look at the actual metrics and independent studies, the "pre-Musk vs. post-Musk" reality is pretty unambiguous. Regarding government censorship, Twitter's pre-Musk transparency reports consistently showed them complying with roughly half of government takedown requests, and they frequently fought overly broad demands in court. Under Musk, data compiled by Rest of World showed that compliance jumped to over 80% (specifically 83% in his first six months), heavily favoring takedown requests from authoritarian-leaning governments. On the topic of algorithmic amplification, y'all argue about whether boosting one side equals censoring the other. Setting the semantics aside, a 2023 Nature study found that X's "For You" algorithm demonstrably amplifies conservative content and steers users toward conservative accounts at a much higher rate than a chronological feed, while actively demoting traditional media. As for moderation and toxicity, the claim that discussing certain topics would automatically get you banned usually ignored that it was generally the manner of the discussion (ie targeted harassment) rather than the topics themselves that triggered enforcement pre-Musk. Post-acquisition, a 2025 PLOS One audit found that measurable hate speech increased by roughly 50%, alongside a significant spike in user engagement with that specific content. Finally, there's the issue of transparency itself. We used to get highly detailed, bi-annual reports that tracked exact volumes of rule enforcement. Those were abruptly paused, and the reports that eventually resumed are heavily stripped down, omitting comprehensive metrics on things like spam and platform manipulation. TL;DR: The data suggests that while you are less likely to get banned by US-centric moderation for controversial cultural takes, the platform is demonstrably more compliant with state-sponsored censorship, less transparent about its operations, and algorithmically tuned to amplify right-leaning content. |